Understanding how we age is one of the challenges researchers face when looking for ways to prevent or treat age-related diseases. Now, artificial intelligence is offering new tools to study this complex process.
A study published in Cell introduces LongevityBench, a set of 17 tasks designed to evaluate how well AI systems can work with questions and data related to ageing. The researchers also developed five AI models specifically trained on ageing research. Despite being much smaller than some general-purpose AI models, they performed as well as or better than larger systems in several of the tests.
The idea is not simply to use AI to predict how old someone is. Researchers hope these specialized models can help them analyse large amounts of biological data, identify patterns and generate new hypotheses about ageing.
One possible application is shown in a separate study published in Nature Biotechnology. Researchers used six different biological clocks, which estimate biological age from proteins in the blood, to investigate the effects of rentosertib, an experimental drug being studied for idiopathic pulmonary fibrosis, a chronic lung disease.
After treatment, all six clocks indicated changes towards a younger biological age in the participants studied. The strongest changes appeared around four weeks after treatment began.
But this does not mean that the drug makes people younger or that it slows ageing. The researchers point out that changes in these measurements could also be related to the treatment of the lung disease itself. More research is needed to determine what the changes detected by the clocks actually mean.
Together, the studies illustrate both the potential and the current limits of using AI to study ageing. AI may help researchers make sense of increasingly complex biological data and identify signals worth investigating. But scientists still need to establish which of those signals genuinely reflect the ageing process.
Rather than giving researchers an answer to how we can stop ageing, these new tools could help them get closer to understanding what is actually happening in our bodies as we age.
References
Zhavoronkov, A., Naumov, V., Sidorenko, D., Aliper, A., Aladinskiy, V., Hasani, R., … & Galkin, F. (2026). An open benchmark and language models for AI in aging biology. Cell, 189(19), 5980-5994.
Zhavoronkov, A., Galkin, F., Chen, S., Ren, F., Aliper, A., Durymanov, M., … & Gladyshev, V. N. (2026). Integration of proteomic aging clocks in a phase 2a clinical trial supports simultaneous geroprotective assessment. Nature Biotechnology, 1-13.
Nature News. (2026). Nature. https://www.nature.com/articles/d41586-026-02913-7
PhD in Sociology from the University of Barcelona. Early Childhood Education Teacher. Substitute Teacher at the Universitat de València.